Onsite LLM Inference Architect for NVIDIA GPU Infra

Delan Associates, Inc; Charlotte (NC); On-site; USD 150,000 - 210,000; Full time

Job summary

Delan Associates, Inc. in Charlotte, NC is seeking an LLM Inference & GPU Systems Consultant to build and maintain on-prem LLM infrastructure on NVIDIA H200 clusters with an OpenShift AI deployment.

The role focuses on production inference, not training, requiring onsite presence 3 days/week and hands-on optimization of GPU workloads, vLLM, TensorRT-LLM, and Hugging Face lifecycle. Candidates should have 8+ years in LLM systems or AI infra, experience with OpenShift AI and RunAI, and strong

Qualifications

• 8+ years of experience as an LLM Systems Engineer or AI Infrastructure Runtime Engineer.
• Hands-on experience with NVIDIA H200 clusters and runtime optimization (KV Cache, prefill/decode).
• Proficiency with OpenShift AI and RunAI for GPU orchestration.
• Experience with modern inference frameworks: vLLM and TensorRT-LLM.
• Experience managing the Hugging Face deployment lifecycle.

Responsibilities

• NVIDIA GPU Runtime Optimization: Drive runtime efficiency and optimize token generation with prefill/decode and KV cache management.
• Inference Serving: Deploy and manage inference engines including vLLM and TensorRT-LLM.
• Hardware Utilization: Optimize GPU throughput, batching, latency; manage RunAI and Kubernetes GPU orchestration.
• Model Lifecycle Management: Oversee Hugging Face model onboarding, deployment, retirement.
• Platform Operations: Operate and maintain the OpenShift AI ecosystem as the primary container platform for GenAI workloads.

Skills

LLM Systems Engineer

AI Infrastructure Runtime

Runtime optimization

GPU orchestration

OpenShift AI

RunAI

vLLM

TensorRT-LLM

Hugging Face lifecycle

Tools

NVIDIA H200 GPUs

OpenShift AI

RunAI

Kubernetes GPU orchestration

vLLM

TensorRT-LLM

Job description

Delan Associates, Inc. in Charlotte, NC is seeking an LLM Inference & GPU Systems Consultant to build and maintain on-prem LLM infrastructure on NVIDIA H200 clusters with an OpenShift AI deployment.

The role focuses on production inference, not training, requiring onsite presence 3 days/week and hands-on optimization of GPU workloads, vLLM, TensorRT-LLM, and Hugging Face lifecycle. Candidates should have 8+ years in LLM systems or AI infra, experience with OpenShift AI and RunAI, and strong